biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V1] Change return type on get_multimodal_embeddings() (#19446)

Browse Source 
Signed-off-by: Russell Bryant <rbryant@redhat.com>
This commit is contained in:
Russell Bryant
2025-06-16 13:32:15 -04:00
committed by GitHub
parent 387bdf0ab9
commit 90f9c2eb5c
37 changed files with 108 additions and 103 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/model_executor/models/paligemma.py
View File

@@ -338,11 +338,11 @@ class PaliGemmaForConditionalGeneration(nn.Module, SupportsMultiModal,
def get_language_model(self) -> torch.nn.Module:
return self.language_model
def get_multimodal_embeddings(
self, **kwargs: object) -> Optional[MultiModalEmbeddings]:
def get_multimodal_embeddings(self,
**kwargs: object) -> MultiModalEmbeddings:
image_input = self._parse_and_validate_image_input(**kwargs)
if image_input is None:
return None
return []
vision_embeddings = self._process_image_input(image_input)
# https://github.com/huggingface/transformers/blob/main/src/transformers/models/paligemma/modeling_paligemma.py#L294 # noqa
vision_embeddings = vision_embeddings * (self.config.hidden_size**-0.5)